Merge 2 pandas data frames on multiple columns

Question

I have 2 dataframes with one of them containing the prediction and actual values for the previous months. As shown below it has prediction from January and February. The values for Febuary are Null.

DF1

Date	Key	Value	prediction
1-1-21	A	33211	22123
1-1-21	BA	43231	32132
1-1-21	C	13431	43432
1-2-21	A	Nan	23421
1-2-21	BA	Nan	44443
1-2-21	C	Nan	32133

The second dataframe has date, key and value for February. I would like to merge them to form a new table with all actual values and a forecasts in one table

DF2

Date	Key	Value
1-2-21	A	33212
1-2-21	BA	52121
1-2-21	C	23123

I want to merge the Values on the key and date into the first table. Final should look like this

Date	Key	Value	prediction
1-1-21	A	33211	22123
1-1-21	BA	43231	32132
1-1-21	C	13431	43432
1-2-21	A	33212	23421
1-2-21	BA	52121	44443
1-2-21	C	23123	32133

I tried pd.merge it creates a new column instead of join into one column and pd.join but it does a left join and drops some of the prediction data. combine first simply appends to the data

Answer 1

这可能是一种在一行（虽然很长）中完成的方法：

df1['Value'] = df1.apply(lambda row: row['Value'] if pd.notna(row['Value']) else df2[(df2['Date'] == row['Date']) & (df2['Key'] == row['Key'])].iloc[0]['Value'], axis=1)